Ati-Driver 8.6/8.7 freeze with KDE 4.1 on logout

[Gregor A.]

Hello, I have more than one problems with my HD4850 and the proprietary ATI drivers.

I have two displays on my card, but I have only one configured.

Only one simple device Section:
Section "Device"
Identifier "Device0"
Driver "fglrx"
BusID "PCI:1:0:0"
Screen 0
EndSection

So I use the clone mode automaticaly.

The problem is:

It is not possible to stop KDM:


/etc/init.d/kdm stop
Stopping K Display Manager: kdm not responding to TERM signal (pid 3494)


Yesterday I tested KDE 4.1, here is it impossible to logout.
If I tray to logout, the complete System is freeze and not respond anymore from network.

This happens only with fglrx and every time.

For me is KDE 4.1 not usable with the proprietary ATI drivers.
It must be possible to logout / change actual user.

More problems I become if I try to use Dualhead mode.
With dualhead is it impossible to change to console. (ctrl + f1) (Catalyst 8.6).

EDIT:
This dualhead problem is solved in catalyst 8.7

@AMD Linux: Please fix this for the next release!

[leodp]

Hi Gregor,

I have a similar problem with an ATI 9700 (single LCD)
I installed kde 4.1 on top of Kubuntu Hardy, and proprietary drivers 8.7

If I log out and in again the screen is completely black, except for the mouse.

It must be a rendering problem, because when I move the mouse above the windows that should be opened at login (I mean: at the position where I expect those windows to be) the mouse changes shape correctly.



I have tried using the free 'ati' driver, but the rendering of windows is veeeeery slow, with some artifacts like black horizontal lines on the desktop background.



Too bad my life with ATI, at each xorg or fglrx release I get better performance (which I do not really use) and new bugs.



Leodp

[energyman]

Hi,
I have seen this too. Luckily I can log out with ctrl-alt-backspace and everything will be fine. Maybe you want to try that?

[ivanovic]

It is not possible to stop KDM:

Remove atieventsd from your start list. Without it is possible to stop Xorg and everything works. This is a problem I have with gentoo, too. Unless you are using a laptop, atieventsd will not do anything anyway, so just run

rc-update del atieventsd

and everything should be fine for you regarding stopping and restarting xorg.

Hi,
I have seen this too. Luckily I can log out with ctrl-alt-backspace and everything will be fine. Maybe you want to try that?

Uhm, this is *not* logging out normally but simply killing xorg. And unless you have a good reason to do so, better don't do it...

[energyman]

If you log out Xorg is killed anyway. All apps are closed. What is the downside of killing X?

Oh, and atieventsd is not in default nor boot. So it can hardly be atieventsd fault.

[Gregor A.]

Remove atieventsd from your start list.

Sorry to inform you, but i have tested this and other Google results already, nothing helps.

Today I tested dualhead (Two "Device" Sections) with Catalyst 8.7 again, everything would be fine, but if I start an game like: ETQW or ut2004, I become problems with screen 0 and screen 1.

I can see in a mess some flicker parts of the game on both screens.

This is not normal.
Normally if I start the game on screen 0, it must run only on screen 0.

@AMD Linux:
Please test the dualhead function better.

I have bought my 4850 to give AMD an chance to stand up.
But at the moment, I am not really happy with this card and fglrx.

[bridgman]

Just a reminder, according to Matthew (if I got it right ) our multihead support is normally single device, not multi-device.

[Gregor A.]

Just a reminder, according to Matthew (if I got it right ) our multihead support is normally single device, not multi-device.

hmmm,

aticonfig --initial=dual-head --screen-layout=above

generates two device sections in my xorg.conf:

Section "Device"
        Identifier  "Device0"
        Driver      "fglrx"
        BusID       "PCI:1:0:0"
EndSection

Section "Device"
        Identifier  "aticonfig-Device[0]-1"
        Driver      "fglrx"
        BusID       "PCI:1:0:0"
        Screen      1
EndSection

[bridgman]

That's not what I expected to see

I'll see what I can find out.

[Gregor A.]

That's not what I expected to see

I'll see what I can find out.

Ok, thanks